Data Viz Principles

Data Viz Resources

Edward Tufte

Tufte: Visual Display of Quantitative Information

William Cleveland

Cleveland: The Elements of Graphing Data

Nathan Yau (FlowingData)

Yau: Visualize This

Telling Stories with Data

One of the best ways to explore and understand a dataset is with visualization.

Telling Stories with Data

What is Statistics?
- hypothesis tests
- pattern finding
- predictive modeling
- storytelling with data can help you solve real-world problems (predicting unrest, decreasing crime) or it can help you stay more informed

Data viz is more than numbers

Journalism

Spending Allocation

Art

Starry night for the color blind

Entertainment

Kobe Bryant Shot Chart

Compelling - Hans Rosling

Hans Rosling

http://www.youtube.com/embed/jbkSRLYSojo?rel=0

Data Viz: What to look for

Patterns

Why so many births around Sept. 25?

Relationships

Age vs. hospital visits

Questionable Data

Fox News

Design Principles

Explain Encodings

what is purple?

Explain Encodings

what is gray?

Label Axes

Calories for menu items

Keep Geometry in Check

proper scaling

Include Sources

source your data

Spotting Visualization Lies

FlowingData Guide for Spotting Visualization Lies:

Types of Graphs

Why use Graphics

Why do you, or have you, in the past used data graphics?
- Exploratory Graphics
- Publication Graphics
- Presentation Graphics

Graphics in R

Visualizing Patterns Over Time

What are we looking for with data over time?
- Trends (increasing/decreasing)
- Are season cycles present?
Identifying these patterns requires looking beyond single points
We are also interested in looking at more the data in more detail
- Are there outliers?
- Do any time periods look out of place?
- Are there spikes or dips?
- What causes any of these irregularities?

Capital BikeShare

Capital Bikeshare Data

bike.data <- read_csv('http://www.math.montana.edu/ahoegh/teaching/stat408/datasets/Bike.csv')

## Parsed with column specification:
## cols(
##   datetime = col_datetime(format = ""),
##   season = col_double(),
##   holiday = col_double(),
##   workingday = col_double(),
##   weather = col_double(),
##   temp = col_double(),
##   atemp = col_double(),
##   humidity = col_double(),
##   windspeed = col_double(),
##   casual = col_double(),
##   registered = col_double(),
##   count = col_double()
## )

Capital Bikeshare Data

bike.data <- bike.data %>% mutate(year = as.factor(year(datetime)), month = as.factor(month(datetime)))
monthly.counts <- bike.data %>% group_by(month) %>% summarize(num_bikes = sum(count), .groups = 'drop') %>% arrange(month)
monthly.counts

## # A tibble: 12 x 2
##    month num_bikes
##    <fct>     <dbl>
##  1 1         79884
##  2 2         99113
##  3 3        133501
##  4 4        167402
##  5 5        200147
##  6 6        220733
##  7 7        214617
##  8 8        213516
##  9 9        212529
## 10 10       207434
## 11 11       176440
## 12 12       160160

Discrete Points: Bar Charts

Discrete Points: Bar Charts - Code

monthly.counts %>% 
  ggplot(aes(y = num_bikes, x = month)) + 
  geom_bar(stat = 'identity') + xlab('Month') + 
  ylab('Bike Rentals') + 
  labs(title = 'Bike Rentals per Month in 2011-2012 \n Capital Bikeshare in Washington, DC', 
       caption = 'Source: www.capitalbikeshare.com')

Discrete Points: Stacked Bar

Discrete Points in Time: Stacked Bar - Code

bike.counts <- aggregate(cbind(bike.data$casual,bike.data$registered),
                         by=list(bike.data$month), sum)
barplot(t(as.matrix(bike.counts[,-1])), 
        names.arg =collect(select(monthly.counts, month))[[1]], 
        xlab='Month', sub ='Source: www.capitalbikeshare.com', 
        ylab='Bike Rentals', 
        main='Bike Rentals per Month in 2011 - 2012 \n Capital Bikeshare in Washington, DC',
        col=c("darkblue","red"),legend.text = c("Casual", "Registered"),
        args.legend = list(x = "topleft"))

Discrete Points in Time: Points

plot(rowSums(bike.counts[,-1])~bike.counts[,1],xlab='Month',
     sub ='Source: www.capitalbikeshare.com', ylab='Bike Rentals', 
     main='Bike Rentals per Month \n Capital Bikeshare in Washington, DC',
     col=c("darkblue"),pch=16,axes=F,
     ylim=c(0,max(rowSums(bike.counts[,-1]))))
axis(2)
axis(1,at=1:12)
box()

Connect the Dots

mean_temp <- bike.data %>% group_by(month) %>%
  summarize(mean_temp = mean(temp),.groups = 'drop') %>% 
  mutate(month = as.numeric(month))

ggplot(aes(y=temp, x= month), data = bike.data) +
  geom_jitter(alpha = .1) + 
  geom_line(inherit.aes = F, aes(y = mean_temp, x = month),
            data = mean_temp, color = 'red', lwd = 2) +
  ylab('Average Temp (C)') + xlab('Month') + 
  labs(title = 'Average Temperature in Washington, DC', 
                       caption = 'Source: www.capitalbikeshare.com')

Visualizing Proportions

What to look for in proportions?
- Generally looking for maximum, minimum, and overall distribution.
Many of the figures we have discussed are useful here as well: for example, stacked bar charts or points to look at changes in proportions over time.
Another possibility, which we will not cover, are plotting with rectangles known as a tree map.

Visualizing Relationships

When considering relationships between variables, what are we looking for?
- If something goes up, do other variables have a positive relationship, negative relationship, or no relationship.
- What is the distribution of your data? (both univariate and multivariate)

Relationships: Scatterplots

Visualizing Relationships: Scatterplots - code

bike.data$tempF <- bike.data$temp * 1.8 + 32
plot(bike.data$count~bike.data$tempF,pch=16,
     col=rgb(100,0,0,10,max=255),ylab='Hourly Bike Rentals',
     xlab='Temp (F)',sub ='Source: www.capitalbikeshare.com',
     main='Hourly Bike Rentals by Temperature')
bike.fit <- loess(count~tempF,bike.data)
temp.seq <- seq(min(bike.data$tempF),max(bike.data$tempF))
lines(predict(bike.fit,temp.seq)~temp.seq,lwd=2)

Visualizing Relationships: Multivariate Scatterplots

pairs(bike.data[,c(12,15,8)])

Multivariate Scatterplots

Relationships: Multivariate Scatterplots

par(mfcol=c(2,2),oma = c(1,0,0,0))
bike.data$tempF <- bike.data$temp * 1.8 + 32
plot(bike.data$count~bike.data$tempF,pch=16,col=rgb(100,0,0,10,max=255),
     ylab='Hourly Bike Rentals',xlab='Temp (F)',
     main='Hourly Bike Rentals by Temperature')
bike.fit <- loess(count~tempF,bike.data)
temp.seq <- seq(min(bike.data$tempF),max(bike.data$tempF))
lines(predict(bike.fit,temp.seq)~temp.seq,lwd=2)

plot(bike.data$count~bike.data$humidity,pch=16,
     col=rgb(100,0,100,10,max=255),
     ylab='Hourly Bike Rentals',xlab='Humidity (%)',
     main='Hourly Bike Rentals by Humidity')
bike.fit <- loess(count~humidity,bike.data)
humidity.seq <- seq(min(bike.data$humidity),max(bike.data$humidity))
lines(predict(bike.fit,humidity.seq)~humidity.seq,lwd=2)

plot(bike.data$count~bike.data$windspeed,pch=16,col=rgb(0,0,100,10,max=255),
     ylab='Hourly Bike Rentals',xlab='Windspeed (MPH)',main='Hourly Bike Rentals by Windspeed')
bike.fit <- loess(count~windspeed,bike.data)
windspeed.seq <- seq(min(bike.data$windspeed),max(bike.data$windspeed))
lines(predict(bike.fit,windspeed.seq)~windspeed.seq,lwd=2)

plot(bike.data$count~as.factor(bike.data$weather),col=rgb(0,100,0,255,max=255),
     ylab='Hourly Bike Rentals',xlab='Weather Conditions',main='Hourly Bike Rentals by Weather')

mtext('Source: www.capitalbikeshare.com', outer = TRUE, cex = .9, side=1)
par(mfcol=c(1,1),oma = c(0,0,0,0))

Relationships: Histograms

hist(bike.data$tempF,prob=T, main='Temperature (F)',col='red',xlab='')

Multiple Histograms

Visualizing Relationships: Multiple Histograms - Code

par(mfrow=c(2,1))
bike.data$reltempF <- bike.data$atemp * 1.8 + 32

hist(bike.data$tempF,prob=T,breaks='FD',
     main='Temperature (F)',col='red',xlab='',
     xlim=c(0,max(c(bike.data$reltempF,bike.data$tempF))))
hist(bike.data$reltempF,prob=T,breaks='FD',
     main='Relative Temperature (F)',col='orange',xlab='',
     xlim=c(0,max(c(bike.data$reltempF,bike.data$tempF))))

Exercise: Visualizing Relationships

Summarize (visually) some relationship from the bike data set